In a conventional fingerprint authentication system, a fingerprint image of a user is first registered in a security token such as a smart card, a universal serial bus (USB) token or the like and then without an outflow information of a registered fingerprint image data of the user, the registered fingerprint image and a target finger print image, i.e., a fingerprint image to be authenticated are compared to each other. The comparison result obtained in the above-mentioned conventional user authentication apparatus is transmitted to outside. This apparatus guarantees higher security than a system in which a central processing unit therein manages fingerprint image data or a system in which registered fingerprint image data is transmitted to an authentication performing circuit in outside thereof. Further, this apparatus can cut off outflow of living body data to thereby protect security of the living body data.
In conventional authentication methods by employing comparison between a registered fingerprint image and a target fingerprint image, methods to estimate similarity between feature points of the registered fingerprint image and those of the target fingerprint image are gaining popularity. For example, in a conventional authentication method, all the feature points extracted from two fingerprint images are checked in view of positions and directions thereof and then similarity between feature points of the registered fingerprint image and those of the target fingerprint image are estimated. The estimated similarity is compared to a predetermined threshold. This method is based on the assumption that the fingerprint image input is carried out in a same position and same direction. Under this assumption, phases of feature points are compared. Accordingly, this method requires that the two fingerprint images should be corrected or amended as ones having same position and direction.
Examples of the methods described in the above are as follows. Namely, there are an article entitled “A real-time matching system for large fingerprint database”, IEEE transactions on pattern analysis and machine intelligence, vol. 18, No. 8, August 1996; Korean Patent Registration NO. 10-0267263 entitled Fingerprint image comparison method free from influence of rotation and transition thereof; and Korean Patent Registration NO.: 10-0198108 entitled “Passing management apparatus by employing fingerprint image and an IC card of a user”.
In detail, in a method described in the article entitled “A real-time matching system for large fingerprint database”, first and second feature points are extracted from registered fingerprint image and target fingerprint image, respectively. Then, the extracted first and second feature points are corrected. When estimating similarity between two fingerprint images by comparing the first and second feature points, the estimation accuracy is enhanced when the two fingerprint images are obtained with a same position and direction.
Since, however, different fingerprint images have generally different positions and directions from each other, a correction procedure for the positions and directions thereof should be added prior to estimation of the similarity. The so-called Hough Transform technique can be applied in the correction procedure to obtain differences in positions and directions.
In detail, a three dimensional region including a bin representing differences of positions and directions is allocated on a memory. Differences of positions and directions of each pair of two feature points are obtained and then accumulated in a corresponding bin within a region. Accumulations for all pairs of two feature points are obtained and then, a bin having a maximum value is obtained as a maximum bin value. Then, correction of positions and directions for the two feature points are carried out with reference to the maximum bin value. However, in this method, since a memory burden greatly increases because a considerable amount of memory is required to allocate the three dimensional region for accumulating differences of positions and directions. Accordingly, this method cannot be applied on a security token having limited hardware resources such as a smart card or a USB token.
Next, the Korean Patent Registration NO. 10-0267263 describes a fingerprint image comparison method free from influence of rotation and transition thereof. This method relates to a fingerprint image comparison method for estimating whether a pair of two fingerprint images is same or not quickly by using feature points data previously extracted from fingerprint images and detecting corresponding points providing information of rotation and transition of the fingerprint images.
In a conventional fingerprint image comparison method, coordinates of all the extracted feature points are checked and accordingly, a lot of calculation burden is entailed in order to check all the candidate feature points having same positions in a space. In the above-mentioned patent, all the feature points are grouped into three parts and a radius of a circumscribed circle for each group, the radius data being stored and utilized in a database. In the patent, when comparing two fingerprint images, the identification between fingerprint images distorted by rotation and/or translation can be checked by selecting only a group including feature points having similar radiuses and quickly checking whether same fingerprint images exist or not. But, in the patent, since there is further needed a memory for storing information of feature point groups. As a result, this method cannot be substantially applied on a security token.
Meanwhile, the Korean Patent Registration NO. 10-0198108 relates to an apparatus for identifying a user by using a fingerprint image and an IC card. In this patent, a user to use a card first registers or stores his or her own fingerprint image. Then, when the user intends to identify himself or herself, a fingerprint image instantly inputted to a card reader is compared with the previously registered fingerprint image in an IC card in a security system. But, the comparison process is performed outside the security system, there entails possibility of outflow of the fingerprint image data.
In view of the foregoing, in the conventional authentication methods mentioned above, there entails problems such as memory burden increase and possibility of outflow of the user's crucial identification data, e.g., fingerprint data as a living body data of the user.